Research map / 3D vision / Human pose estimation
ZoomNAS: Searching for Whole-body Human Pose Estimation in the Wild
Lumin Xu, Sheng Jin, Wentao Liu, Chen Qian, Wanli Ouyang, Ping Luo, Xiaogang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add searched connections (feature stage, RoI expansion) between BodyNet and heads (ablation) | ZoomNAS · COCO-WholeBody V1.0 (whole-body) | AP: 63.9 -> 65.4 | Improved |
| add searched connections (ablation) | ZoomNAS · COCO-WholeBody V1.0 (face) | AP: 84.6 -> 88.9 | Improved |
| add searched connections (ablation) | ZoomNAS · COCO-WholeBody V1.0 (hand) | AP: 58.3 -> 62.5 | Improved |
| automatic computation allocation instead of proportional allocation (ablation) | ZoomNAS · COCO-WholeBody V1.0 (whole-body) | AP: 62.0 -> 65.4 | Improved |
| automatic computation allocation instead of proportional allocation (ablation) | ZoomNAS · COCO-WholeBody V1.0 (foot) | AP: 45.2 -> 61.7 | Improved |
| automatic computation allocation instead of proportional allocation (ablation) | ZoomNAS · COCO-WholeBody V1.0 (face) | AP: 91.3 -> 88.9 | Hurt |
| joint training instead of successive training (ablation) | ZoomNAS · COCO-WholeBody V1.0 (whole-body) | AP: 65.2 -> 65.4 | No clear effect |
| zoom-in face/hand heads on top of HRNet backbone (ZoomNet vs HRNet) | ZoomNet · COCO-WholeBody V1.0 (whole-body) | AP: 58.6 -> 63.0 | Improved |
| neural architecture search over ZoomNet (ZoomNAS vs ZoomNet) | ZoomNet · COCO-WholeBody V1.0 (whole-body) | AP: 63.0 -> 65.4 | Improved |
| neural architecture search over ZoomNet (ZoomNAS vs ZoomNet) | ZoomNet · COCO-WholeBody V1.0 (whole-body) | GFLOPs: 28.49 -> 18.02 | Improved |
| pretrain on WholeBody-Face | HRNetV2-W18 · WFLW | NME: 4.60 -> 4.01 | Improved |
| pretrain on WholeBody-Face | HRNetV2-W18 · COFW | NME: 3.45 -> 3.36 | Improved |
Later papers that compared with this method
3 later accepted papers report ZoomNAS as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Spatial-Aware Regression for Keypoint Localization (CVPR 2024 · better on 2 of 2 shared results)
- Sapiens: Foundation for Human Vision Models (ECCV 2024 · better on 1 of 1 shared result)
- GTPT: Group-based Token Pruning Transformer for Efficient Human Pose Estimation (ECCV 2024 · better on 0 of 1 shared result)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- OpenPose — OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields (TPAMI 2019) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Whole-Body Human Pose Estimation in the Wild (ECCV 2020 · similarity 0.95)
- Single-Network Whole-Body Pose Estimation (ICCV 2019 · similarity 0.91)
- DOPE: Distillation Of Part Experts for whole-body 3D pose estimation in the wild (ECCV 2020 · similarity 0.90)
- H3WB: Human3.6M 3D WholeBody Dataset and Benchmark (ICCV 2023 · similarity 0.88)
- Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · similarity 0.86)
- PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
